Reverse genetics of salmonid alphavirus
The first infectious full-length cDNA of salmonid alphavirus (SAV2) was generated by Moriette et al. in 2006. A hammerhead ribozyme sequence placed immediately upstream of viral 5’UTR and downstream of CMV or T7 promoter functions as a self-cleavage enzyme which has shown to be crucial for the recovery of recombinant SAV2 (Moriette et al., 2006).
A similar strategy has been applied to generate recombinant SAV3 (Karlsen et al., 2009;
Karlsen et al., 2010; Paper III in this thesis), despite different backbone vectors were used to carry the viral genome. In the present study, we showed that rSAV3 can also be rescued from a full-length cDNA construct without incorporating the HH sequence at 5’end, although the recovery efficiency was significantly lower for this construct compared to HH construct (Paper IV). The explanation might be that non-HH cDNA construct contains a non-viral sequence between CMV promoter and 5’ UTR (Paper III) which could impede the viral replication. If this is the case, removing the non-viral sequence at 5’ end should improve virus recovery. However, it has been shown that recombinant SAV2 could not be recovered without a space sequence between the promoter and the 5’UTR (Leberre et al., 2011). The mechanism behind this remains unknown.
The recovered recombinant SAV2 (rSDV) was subjected to in vivo infection in juvenile rainbow trout and demonstrated to be fully attenuated and protective against subsequent challenge. Interestingly, the protective effect is seen only when the virus is recovered from CHSE-214 cells cultured at 10°C but not from BF-2 cells cultured at 14°C (Moriette et al., 2006). Further genome sequencing revealed six amino acid changes in the structural polyprotein (three in the E2, two in the 6K, and one in the E1). In this thesis, we demonstrated infectivity of rSAV3 in vitro and in addition we have also assessed the in vivo effect by intramuscular injection of full-length cDNA vectors into fish. At 8 weeks post injection, the target organs (pancreas and heart) were examined and shown to exhibit classical pathological changes indicating that pathogenic rSAV3 is formed in vivo. Mortality was not recorded (unpublished data) in line with other studies. To our knowledge, this is the first time that recombinant SAV was recovered from injecting cDNA into fish. The propagated virus in fish was reisolated from heart tissue and inoculated onto CHSE-214 cells where CPE was observed at 6 days post inoculation. Whether reducing the injection dose would reduce
37
pathological changes in internal organs and at the same time conferring protective immunity in fish remains to be evaluated.
Host-pathogen interaction in vivo
Characterization of immune gene expression post challenge is important for understanding mechanisms underlying host-pathogen interactions and for vaccine development. In paper I, fish were challenged with the SAV3-H10 isolate and then assessed for pathological changes, viral load, and expression of immune genes in different target organs over the course of infection. Pathological changes were first found in the pancreas and subsequently in the heart, in line with what has been described previously during natural infections (McLoughlin &
Graham, 2007). Over the same period, IFN and ISGs were upregulated in association with the increase in viral load. It is noteworthy that ISG15 was upregulated 650-fold at two weeks post infection, before any pathological changes were evident. In conclusion, the results shed light on the importance of IFN system during SAV3 infection yet its antiviral role remains obscure as highly up-regulated IFN and ISGs in fish did not successfully hinder virus infection. This therefore brought us into the second study to investigate the role of IFN on viral replication (Paper II).
SAV and susceptibility to IFN system
It has been shown for other alphaviruses that IFN-α plays a pivotal role in protection against infection by limiting viral replication (Aguilar et al., 2005;Grieder & Vogel, 1999;Stanton et al., 1989). In agreement with this, our study showed that SAV3 is highly sensitive to IFN-α (Paper II), as cells pretreated with recombinant IFN-α before virus infection exhibited anti-SAV3 property. However, when virus infection and viral genome translation/replication occurred prior to treatment with recombinant IFN, the induced responses come too late to prevent viral replication and production of infectious progeny. This is in line with the observation that endogenous IFN increases concurrently with increase in viral RNAs copies, shown both in vitro and in vivo (Paper I, II, and III) but this is insufficient to hinder successful production of infectious progeny. In spite of this, results from Paper III and Paper IV suggest that IFN response regulates viral RNA replication to some extent. This was evidenced by early development of CPE and higher end-point virus titers when virus was propagated in CHH-1 cells (low competent) compared to CHSE-214 and in BF-2 cells (high IFN-competent). A similar study conducted previously on SAV1 infection in CHH-1 cells also showed earlier CPE compared with CHSE-214 and SHK-1 cells, yet no difference was found
38
with regard to the virus titers between these cell lines (Herath et al., 2009). The discrepancy with our results might possibly be due to in vitro virulence differences between subtypes of SAV. The reduced IFN-α and ISGs (Mx and ISG15) response to SAV3 infection in CHH-1 cells could offer a superior environment for viral replication and assembly, which is further supported by the higher level of heterologous antigen expressed by the SAV3 replicon vector in CHH-1 cells compared to CHSE-214 and BF-2 cells (Paper IV). Taken together, these results suggest that CHH-1 cells might be well suited for generation of high titer SAV3 for inactivated vaccine development. Similarly, an IFN-deficient cell line, Baby hamster kidney-21 (BHK-kidney-21), has been widely used for the production of high titer alphaviruses (Atkins, 1979).
In paper II, we showed that recombinant salmon IFN-γ exhibited marginal antiviral effect against SAV3. In contrast to our result, a comparable study (Sun et al., 2011) suggested that recombinant trout IFN-γ had antiviral effect in vitro. Several factors might provide explanations for the differences seen. First, the amino acid sequence differs, around 10%
between salmon and trout IFN-γ, and this might affect biological functions. This might be of importance since rainbow trout recombinant IFN-γ was used in the study by Sun et al (2011) and their main results for SAV3 are based on infecting TO cells (derived from Atlantic salmon). Furthermore, the temperature for IPTG induction (28°C in Sun et al’s study and 37°C in our study) and purification of IFN-γ from either soluble or insoluble phase might also influence on the function of purified proteins. Last, the SAV3 strains used in two studies possess different in vitro virulence. The strain used in the study by Sun et al. was non-lytic and they only see a 4 log10 increase (by real-time PCR) by day 14 post infection without any CPE. The strain we used is lytic and causing CPE at 8 days post infection in TO cells. On this basis it is questionable if the results are comparable. It should also be added that the authors conclude that the observed effects of IFN-γ on SAV3 replication is also partly dependent on IFNa induction (Sun et al. 2010).
Recombinant virus recovered from DNA and RNA based methods – impact from IFN susceptibility
In Paper III, we recovered rSAV3 from a full-length cDNA (pSAV3-HHFL) clone and the generated virus titre was comparable to what was found for wild-type virus. Furthermore, the rescued recombinant SAV3 (rSAV3) caused similar plaque morphology in infected cells as wild-type virus (data not shown). These results show the infectivity of rescued rSAV3 (plasmid-based) was identical to wild-type virus. A different approach, through transfection of
39
cells with in vitro transcribed RNAs, has been applied for rescue of other recombinant alphaviruses, like SFV and SINV (Liljestrom & Garoff, 1991a;Migliaccio et al., 1989). The same approach was successfully tested for recovery of rSAV3 in CHSE-214 cells. However, despite the in vitro transcribed RNAs were confirmed to be intact prior to transfection, expression of viral proteins in cells transfected with in vitro transcribed RNAs was much lower compared to cells transfected with the infectious cDNA clone (Figure 11A). Further investigation revealed an immediate and strong induction of IFN-α and ISGs responses following transfection of in vitro transcribed RNA into cells (Figure 11B). In contrast, transfecting CHSE cells with the full-length cDNA clone did not provoke such an anti-viral response (Figure 11B), which likely explains the differences. Massive uptake of in vitro transcribed RNAs into cells is likely recognized by single-stranded viral RNA sensors, such as TLR7 and TLR8 (in endosomes) and RIG-I in the cytosol (Diebold, 2010;Dixit & Kagan, 2013;Koyama et al., 2008;Yoneyama & Fujita, 2010), with subsequent IFN induction. This is in line with the results from Paper II, where the competition between the initiation of viral RNA replication and the induction of IFN response is critical for establishment of a successful infection. We further observed that virus recovery was augmented when in vitro transcribed RNAs were transfected into CHH-1 cells, a cell line with limited IFN expression (data not shown). In this regard, finding a SAV3 susceptible cell line completely lacking IFN expression (analogue to BHK-21 cells used for other alphaviruses) can facilitate virus recovery from the in vitro transcribed RNA.
40
cDNA RNA
IFN
24h 48h 72h 96h 24h 48h 72h 96h 0.5
24h 48h 72h 96h 24h 48h 72h 96h 0.1
24h 48h 72h 96h 24h 48h 72h 96h 0.1 live-attenuated vaccines, which is expected to elicit both the humoral- and cell-mediated immunity, and hence conferring improved protection compared to inactivated or recombinant subunit vaccines. Deleting the entire 6K gene from SFV leads to successful production of attenuated virus. In contrast, complete deletion of 6K gene from the SAV3 cDNA construct (pSAV3-HHΔ6K cDNA) did not generate infectious particles (Paper III) and were therefore not a suitable strategy for developing an attenuated viral vaccine. Despite this, expression of viral proteins was detected, suggesting the RNA replication and translation were not impaired.
Viral protein was however not detected on the cell surface, implying that translated viral proteins were somehow retained in the ER or Golgi and not transported to the cell membrane.
(A)
(B)
Figure 11. (A) Viral protein expression in CHSE-214 cells transfected with SAV3 full-length cDNA clone (Left) or in vitro transcribed RNAs generated from the same clone (Right). (B) The immune response in CHSE-214 cells elicited by transfection of either infectious cDNA or in vitro transcribed RNAs.
41
This can possibly be caused by an error in polyprotein cleavage/processing. However, to confirm this will require antibodies against each single viral protein and another challenge is the limited amount of viral protein expressed in pSAV3-HHΔ6K cDNA transfected cells, likely due to low transfection rate and non-spreading virus production. The use of a cell line with intact IFN production could also be an obstacle for viral protein expression. Nevertheless, the pull-down experiment revealed a larger size E2 protein produced in cells transfected with pSAV3-HHΔ6K cDNA (Paper III). Whether this is caused by the defect of polyprotein cleavage remains to be determined, but we cannot exclude the possibility of the erroneous post-translational modification on E2 protein resulting in a larger molecular weight. Another strategy to generate an attenuated variant could be the construction of a cDNA clone with partially deleted 6K gene preserving a few nucleotides on both ends which are thought to provide cleavage sites. In addition, a unique seven-amino acid sequence (GVRGWSA) was identified in the 6K protein of SAV subtypes 1-3, not found in other alphaviruses, possibly owing to fish are poikilothermic and also live at low water temperatures, leading to dissimilar lipid composition in the cell membrane. This unique sequence in SAV 6K therefore constitutes an interesting target for the future work with regard to its functional role and the potential discovery of an attenuated virus variant.
SAV3 RNA recombination is documented in vitro
Viral RNA recombination is important in evolution of viruses. Despite RNA recombination has been shown for other alphaviruses, to what extent this occurs for SAV remains unclear. A recent study suggested that RNA recombination might appear in vivo during SAV infection (Petterson et al., 2013), as the authors detected various deletions within viral genomes from different field isolates. In paper III, we demonstrated that RNA recombination occurred in cells co-transfected with pSAV3-HHΔ6K cDNA and helper cDNA constructs containing complete structural genes. CPE was found in 14 out of 48 co-transfected wells indicating that restoration of a fully infectious virus occurred at a relatively high frequency. Further analysis of the cross-over site in more detail in one well with CPE revealed that several different recombinations had taken place at slightly different sites. This was evidenced by different restriction patterns when recombinant RNAs were RT-PCR amplified, cloned, and digested with restriction enzymes. However, after one round of virus passage in cells, restriction fragment analysis of the cross-over region became homogenous (Paper III), likely due to the selection of viable viral particles/viral clones able to infect cells in the next round of infection.
It is expected that also non-viable truncated viral genomes generated in transfected cells can
42
be packaged into viral particles and passaged, and therefore a small amount of deletion variants might co-exists with full-length variants. It is however difficult to decide which of the generated deletion variants are viable. This could be confirmed by plaque purification and genome sequencing in future studies. Homologous analysis of full genome of SAV subtypes 1-3 revealed differences with regard to deletions/additions, which might be explained by RNA recombination occurring between subtypes. This remains a speculation and is (likely) in contrast to a recent study by Karlsen and co-workers (2013) showing that the six subtypes of SAV represent independent introductions to farmed fish populations. This is based on six full-genome sequence analyses and 71 partial sequencing of the structural ORF, which suggest that all six subtypes diverged prior to the twentieth century before rainbow trout was introduced into European aquaculture. The interpretation is that the different subtypes must have existed in wild populations or a reservoir. The ancestors of the strains found in aquaculture today thus likely represent independent introductions to farmed fish populations from the wild reservoir and each of the subtypes has developed into self-sustainable epizootics (Karlsen et al. 2013). As mentioned above, recombination is likely to occur in vivo (Petterson et al. 2013) and the possibility for mixed infections (with different subtypes) in the same population of fish cannot be ruled out. Recently, marine SAV2 has been found in areas (in Norway) where SAV3 is endemic (own observations). The molecular mechanism of RNA recombination for SAV remains to be understood, but the results presented in Paper III has delivered an important message regarding the safety of vaccines containing full-length or partially deleted viral genomes. There is a risk of RNA recombination between vaccine virus and field virus leading to reversion to virulence.
SAV3 Replicon based vaccine
The construction of SAV2 replicon containing all the non-structural genes plus GFP or Luciferase reporter gene was described previously by Moriette et al., 2006. Later, a similar replicon construct was developed for SAV3 (Karlsen et al., 2009) and demonstrated to be functional in fish, shrimp and mammalian cells in a temperature range of 4-37°C (Olsen et al., 2013), suggesting that despite the progeny of SAV is produced only at low temperature (10-15˚C), the replication machinery is not temperature restricted. This is supported by a recent study showing that RNA replication occurred at both 12 ˚C and 20 ˚C and further demonstrating that E2 glycoprotein is responsible for this low-temperature dependency and requires co-expressed E1 for virion formation (Hikke et al., 2014). Furthermore, SAV3 replicon encoding the infectious salmon anemia virus (ISAV) hemagglutinin-esterase (HE) is
43
proven to be an efficacious vaccine against ISA in fish (Wolf et al., 2013) using intramuscular immunization but no protection was induced when pSAV/HE was injected intraperitoneally (Wolf et al., 2014). Nevertheless, in contrast to other alphaviruses showing that the self-replicating RNA molecules (replicon) is capable of expressing its gene of interest to a level as high as 20% of total cellular protein production (Pushko et al., 1997), the level of antigen expression encoded by SAV replicon was in general low (Olsen et al., 2013). As antigen dose correlates with protection, augmentation of the protein level expressed by a vaccine vector would be beneficial in vaccination. In Paper IV, we optimized antigen expression levels by incorporating a translational enhancer (the N-terminal 102 nucleotides of the capsid gene) into the expression vector. The expression level of a reporter gene, EGFP, in one cell unit was measured by flow cytometer and the result suggested this modified replicon vector containing the translational enhancer significantly enhanced antigen expression in the tested cell line (CHH-1). We also confirm that this vector was able to express the complete structural genes of IPNV (segment A) by examining expression of IPNV proteins in transfected cells in vitro.
A recent study shows when Atlantic salmon was immunized with SAV3 replicon expressing IPNV segment A polyprotein and further challenged by cohabitation with IPNV shedder fish, low to moderate protection was obtained (Abdullah et al., 2015). However, it might still be interesting to evaluate whether the enhanced antigen expression obtained by the replicon generated in this study can improve the protection against IPNV.
In addition to the replicon-based vaccine, suicide viral particles, which undergo only one round of infection have been developed to be used as potential vaccines in the future. These particles go through a natural infection route in the host and replicate in the target sites (Zhou et al., 1994). We have attempted to make suicide particles of SAV3, however without success.
Co-transfection of a SAV3 replicon (carrying the EFGP gene) and a helper cDNA construct containing the complete structural genes did not lead to packaging of viral particles. Despite EFGP was expressed in transfected cells, CPE was not observed, and when supernatant from transfected cells was passaged to new cells, EGFP expression was no longer detected. It remains uncertain why SAV3 replicon together with helper cDNA was not able to form infectious virus particles as shown for other alphaviruses. One possible explanation might be the location of the packaging signal. For SFV, it is known that virus package signals are found within the nsP2 gene and for SINV it is found within nsP1 (Kim et al., 2011;Kim et al., 2013).
If the packaging signal of SAV is found outside the non-structural genes, then capsid protein will not be able to pack the replicon. It is also likely that the transfected permissive cell lines
44
used in this study (CHSE-214 and CHH-1 cells) possess an interferon response sufficient to block the infection process if the recombinant virus is attenuated compared to wild-type virus.